[ovs-dev] [PATCH 2/2] timeval: Recover from failed timer_create() calls.

Ben Pfaff blp at nicira.com
Fri Oct 5 17:13:49 UTC 2012


On Thu, Oct 04, 2012 at 07:15:43PM -0700, Ethan Jackson wrote:
> The timer_create() system call is not supported in ESX and returns
> an error when called.  Aborting when this system call fails seems a
> bit extreme. So instead, this patch simply falls back to disabling
> the cached time optimization.
> 
> Signed-off-by: Ethan Jackson <ethan at nicira.com>

There are a couple of uses of CACHE_TIME outside timeval.c, in the
unit tests.  It would be better, I think, if we exported the correct
runtime value.  So I would suggest that we move CACHE_TIME into
timeval.c and add a time_cached() or similar function.

The comment on cache_time is kind of hard to understand.  I'd make it
more explicit, e.g.:

    True on systems (particularly Linux on x86-64) where clock_gettime()
    is cheap.  On these systems, we don't bother caching the current time
    but check it on every call to e.g. time_msec().

    False on systems where clock_gettime() is relatively expensive.  On
    these systems we cache the current time and set up a periodic SIGALRM
    to remind us to check it again.

    Also false on systems (e.g. ESX) that don't support setting up timers
    based on a monotonically increasing clock.



More information about the dev mailing list